Chapter Assessment: Statistics for Evals

This review checks the statistics behind every eval result: how sure a pass rate is, how many cases a decision needs, and how to compare versions honestly. The bank holds 27 questions, three for each lesson, and each attempt draws ten. Most questions give a result from the job ticket assistant and ask what it supports, and several ask you to compute a number from the lesson's own inputs. Keep four things in mind: the standard error formula, the Wilson interval near 100 percent, McNemar's test on fixed and broken cases, and the effect of many comparisons on a winning variant